The most impressive thing you can do on Day 1 of a consulting engagement is not a slide deck. It’s producing in two hours what the client expected to take two weeks.
I’m about to start at a bank where one workstream is regulatory gap analysis — comparing new AI regulations against the bank’s existing internal policies. The client’s team does this manually: read the regulation, read the internal standard, identify gaps, write up findings. It takes weeks per regulation, across multiple jurisdictions.
Before I start, I’m building the machine. Not a prototype. Not a concept. A working tool that:
- Accepts any regulation (HKMA, MAS, EU AI Act — already loaded)
- Accepts any internal policy document (upload via API)
- Decomposes the regulation into discrete requirements with citations
- Compares each requirement against the policy baseline
- Classifies as Full / Partial / Gap with paragraph-level evidence
- Runs a completeness audit to catch missed requirements
The tool exists. The corpus is loaded. The prompts are tuned for AI governance specifically.
On Day 1, I won’t pitch this. I’ll listen, diagnose, understand what they actually have. But when the moment comes — “we need to assess against MAS AIRM Guidelines” — I’ll ask for their internal standard and deliver results the same afternoon.
The prep investment is small. The signal it sends is large: this person doesn’t just understand the problem, they’ve already solved part of it.
The general principle: if you know what a client needs before you arrive, build the tool, not the presentation. Ideas are cheap. Working systems that produce results are not.